© Artificial Intelligence?
Monday, December 10, 2018
When ‘Edmond de Belamy’ was auctioned by Christie’s in October, nobody asked whether copyright can rest on a portrait made by an algorithm. The main question at the auction was how much the portrait would be worth, knowing that it was not made by a famous painter, but by a computer.
The fact that this does not affect the value was made clear at the auction when a prize of $ 432,000 was paid for the painting of “the hand of” an algorithm by the artist collective ‘Obvious’.
The painting portrays ‘Edmond de Belamy, from the Belamy family’. The portrait is part of a collection of eleven portraits of the fictional family Belamy, which were made by Artificial Intelligence. La Comptesse de Belamy (pictured here) is part of the eleven portraits of the Belamy family.
How exactly did the process of making these portraits work? The algorithm is ‘fed’ with data, in this case 15,000 portraits from the 14th to the 20th century. The dual nature of the algorithm enables the algorithm to make newportraits on the basis of the data and to check for itself whether the portrait made can be distinguished from a portrait made by a human being. The goal of the art collective Obvious was to have the algorithm make new portraits that resemble man-made portraits from the 14th to the 20th century in a way that the algorithm itself would no longer be able to make this distinction. And with success; the Belamy portraits have been individually identified as a manmade portraits by the algorithm.
The series Belamy portraits raises various copyright questions. First of all: are the portraits subject to copyright protection? After all, copyright protection requires that the portrait has its own original character and bears the personal stamp of the maker (EOK & PS). In order to fulfill the requirement of its own original character, the portrait may not be borrowed from another work. It is necessary that the portraits take sufficient distance from earlier works, including those that have been entered as data.
This requires the portrait to be the result of creative choices made by the maker (personal stamp). But who should be regarded as the maker in this case, and in this case can creativity or creative choices be discussed at all? Strictly speaking, the algorithm has made the portraits, but a computer cannot be regarded as a maker in the sense of the Copyright Act. This is because a copyrighted work must be the result of creative choices of the human mind (and a computer is a ‘matter’ without legal personality!). Perhaps the art collective Obvious can be regarded as a maker, because with a certain vision she has selected the data and entered it into the algorithm. According to this reasoning, the algorithm would only be a machine that implemented the vision of the art collective, such as a camera that does for the photographer. In order for this argument to succeed, the art collective must make enough creative choices, for example with selecting and entering the data.
If the creative contribution of the art collective proves to be so minimal that the algorithm appears to function autonomously, it becomes difficult to claim copyright protection. Only the claim of the writer of the software of the algorithm could still succeed if he can argue that the portraits must be regarded as a direct result of the software to which he has copyright. If the art collective Obvious has the rights to the software of the algorithm, they could claim to be the owner of any copyrights on the portraits.
For the appreciation of the painting at the auction, all these questions were of no importance. But, what to do with an identical copy of one of the Belamy’s? In order to speak of infringement, you must first speak of a copyrighted work and that is not obvious in this case.
Moïra Truijens, counsel & Lisanne Steenbergen, paralegal
Print this article